Skip to contentSkip to conversation

OMAP 3 ISP and N900 sensor driver update

DISCUSSION Sakari Ailus 5 participants 9 messages Nov 6, 2009 – Nov 11, 2009 Media (V4L/DVB)

lore
Share & tools
lore
#0 lore raw

Hi,

I have updated the OMAP 3 ISP driver in Gitorious:

<URL:http://www.gitorious.org/omap3camera>

Major changes since the last update:

- The Nokia N900 (aka rx-51) sensor drivers are available (will be posted to the list shortly)
- Say goodbye to v4l2-int-device, welcome the v4l2_subdevice interface (thanks to Laurent Pinchart)
- Miscellaneous stability fixes and cleanups
- H3A rework (by David Cohen)
- Resizer rework (by Antti Koskipää)

The next task is then the moving to Media controller, I guess.

Cheers,

Signature · 2 lines
Sakari Ailus
[email protected]
#1 lore raw
Quoted 18 lines

-----Original Message-----
From: [email protected] [mailto:linux-media- [email protected]] On Behalf Of Sakari Ailus
Sent: Friday, November 06, 2009 6:22 PM
To: [email protected]
Cc: Laurent Pinchart; Hans Verkuil; Cohen David Abraham; Koskipää Antti Jussi Petteri; Toivonen Tuukka Olli Artturi; Zutshi Vimarsh (Nokia-D-MSW/Helsinki); [email protected]; Aguirre, Sergio; Ivan Ivanov; Stan Varbanov; Valeri Ivanov; Atanas Filipov
Subject: OMAP 3 ISP and N900 sensor driver update

Hi,

I have updated the OMAP 3 ISP driver in Gitorious:

<URL:http://www.gitorious.org/omap3camera>

Quoted 7 lines

Major changes since the last update:

- The Nokia N900 (aka rx-51) sensor drivers are available (will be posted to the list shortly)
- Say goodbye to v4l2-int-device, welcome the v4l2_subdevice interface
(thanks to Laurent Pinchart)

[Hiremath, Vaibhav] Thanks Sakari and Laurent, this will definitely unblock me.

Thanks,
Vaibhav

Quoted 16 lines

- Miscellaneous stability fixes and cleanups
- H3A rework (by David Cohen)
- Resizer rework (by Antti Koskipää)

The next task is then the moving to Media controller, I guess.

Cheers,

--
Sakari Ailus
[email protected]
--
To unsubscribe from this list: send the line "unsubscribe linux- media" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#2 lore raw
In reply to Sakari Ailus

Sakari,

Thanks for the update...

Who is working on the CCDC driver for OMAP35xx? After a week or so,
I need to start migrating the CCDC driver to sub device interface so that application can directly configure the parameters with out having to go through video node. Ultimately it is expected that ccdc will have a
device node that will allow application to open the device and configure
the parameters (in the context of Media controller). But to begin with
I intend to port the existing CCDC driver for DM6446 and DM355 to
sub device interface. Since the VPFE IPs are common across DM6446 &
OMAP 35xx, we could use a common sub device across both platforms.

So I this context, could you please update me on the CCDC development on OMAP platform that you work?

Murali Karicheri
Software Design Engineer
Texas Instruments Inc.
Germantown, MD 20874
email: [email protected]

Quoted 38 lines

-----Original Message-----
From: [email protected] [mailto:linux-media- [email protected]] On Behalf Of Sakari Ailus
Sent: Friday, November 06, 2009 7:52 AM
To: [email protected]
Cc: Laurent Pinchart; Hans Verkuil; Cohen David Abraham; Koskipää Antti Jussi Petteri; Toivonen Tuukka Olli Artturi; Zutshi Vimarsh (Nokia-D- MSW/Helsinki); [email protected]; Aguirre, Sergio; Ivan Ivanov; Stan Varbanov; Valeri Ivanov; Atanas Filipov
Subject: OMAP 3 ISP and N900 sensor driver update

Hi,

I have updated the OMAP 3 ISP driver in Gitorious:

<URL:http://www.gitorious.org/omap3camera>

Major changes since the last update:

- The Nokia N900 (aka rx-51) sensor drivers are available (will be posted to the list shortly)
- Say goodbye to v4l2-int-device, welcome the v4l2_subdevice interface (thanks to Laurent Pinchart)
- Miscellaneous stability fixes and cleanups
- H3A rework (by David Cohen)
- Resizer rework (by Antti Koskipää)

The next task is then the moving to Media controller, I guess.

Cheers,

--
Sakari Ailus
[email protected]
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#3 lore raw

Hi Murali,

Quoted 5 lines · Muralidharan

Sakari,

Thanks for the update...

Who is working on the CCDC driver for OMAP35xx?

Just for the sake of correctness, we're working on an OMAP34xx, not an OMAP35xx. I don't think that makes a big difference though.

As far as I know nobody on our side is currently working on the CCDC driver. We're focusing on the previewer and resizer first.

Quoted 8 lines

After a week or so, I need to start migrating the CCDC driver to sub device interface so that application can directly configure the parameters with out having to go through video node. Ultimately it is expected that ccdc will have a device node that will allow application to open the device and configure the parameters (in the context of Media controller). But to begin with I intend to port the existing CCDC driver for DM6446 and DM355 to sub device interface. Since the VPFE IPs are common across DM6446 & OMAP 35xx, we could use a common sub device across both platforms.

Coordinating our efforts on that front would indeed be very nice.

Quoted 2 lines

So I this context, could you please update me on the CCDC development on OMAP platform that you work?

I haven't checked the Davinci VPFE drivers recently. I suppose they already use the v4l2_subdev interface for their I2C sensors and tuners. If not, that would be the first step.

On the OMAP34xx platform, the ISP driver is already somehow separated from the omap34xxcam driver, although not nicely. In a nutshell, here's the current plan. Parts 1 and 2 are already implemented and code is available in Sakari's linux-omap tree.

1. Board code registers an omap34xxcam platform device. The platform data contains an array of v4l2_subdev_i2c_board_info structures filled with information about all I2C sub-devices (sensor, flash controller, lens controller, ...).

2. The omap34xxcam driver is loaded. Its probe function is called with a pointer to the platform device. The driver registers a v4l2_device and creates I2C subdevices using the data supplied through platform data.

3. The omap34xxcam driver calls the ISP core with a pointer to the v4l2_device structure to register all ISP subdevices. The ISP core maintains pointers to the ISP subdevices.

As ISP submodules (CCDC, previewer, resizer) are not truly independent, we were not planning to split them into separate kernel modules. The ISP core needs to call explicitly into the submodules for instance to dispatch interrupts.

It should be possible to use a single CCDC code base across multiple platforms. The ISP core module would depend on the CCDC module directly. I'm not sure how the CCDC module should be called though. An omap prefix won't work, as it's used on Davinci as well, and an ISP prefix is too generic. Do you have any internal code name for the ISP device used on Davinci and OMAP platforms ?

The code will be pushed to linux-omap when available, but I can't commit to any deadline. If you start working on the Davinci driver before we post anything, could you please post patches and/or RFCs on the linux-media mailing list (and CC'ing us) ? Confronting our ideas as soon as possible will (hopefully) avoid diverging too much in our implementations.

Signature · 3 lines
Regards,

Laurent Pinchart
#4 lore raw

Hi Laurent,

I have also copied Vaibhav from TI in this email since this is of interesting to his area of work as well.

Murali Karicheri
Software Design Engineer
Texas Instruments Inc.
Germantown, MD 20874
email: [email protected]

Quoted 6 lines
Quoted 2 lines

Who is working on the CCDC driver for OMAP35xx?

Just for the sake of correctness, we're working on an OMAP34xx, not an OMAP35xx. I don't think that makes a big difference though.

That is right!

Quoted 10 lines

As far as I know nobody on our side is currently working on the CCDC driver. We're focusing on the previewer and resizer first.

Quoted 1 line

After a week or so, I need to start migrating the CCDC driver to sub

device

Quoted 1 line

interface so that application can directly configure the parameters with

out

Quoted 3 lines

having to go through video node. Ultimately it is expected that ccdc will have a device node that will allow application to open the device and configure the parameters (in the context of Media controller). But to

Quoted 8 lines

begin

Quoted 1 line

with I intend to port the existing CCDC driver for DM6446 and DM355 to

sub

Quoted 2 lines

device interface. Since the VPFE IPs are common across DM6446 & OMAP 35xx, we could use a common sub device across both platforms.

Coordinating our efforts on that front would indeed be very nice.

Ok.

Quoted 7 lines
Quoted 2 lines

So I this context, could you please update me on the CCDC development on OMAP platform that you work?

I haven't checked the Davinci VPFE drivers recently. I suppose they already use the v4l2_subdev interface for their I2C sensors and tuners. If not, that
would be the first step.

Yes.

The vpfe drivers defines all sub device configuration data in the platform specific code and use that information to load up sub devices. Example tvp514x is loaded by vpfe capture driver and also mt9t031.

Quoted 8 lines

On the OMAP34xx platform, the ISP driver is already somehow separated from the
omap34xxcam driver, although not nicely. In a nutshell, here's the current plan. Parts 1 and 2 are already implemented and code is available in Sakari's
linux-omap tree.

I will check this out some time next week as I am currently tied up with internal release deadline.

Quoted 5 lines

1. Board code registers an omap34xxcam platform device. The platform data contains an array of v4l2_subdev_i2c_board_info structures filled with information about all I2C sub-devices (sensor, flash controller, lens controller, ...).

vpfe capture does similar things.

Quoted 5 lines

2. The omap34xxcam driver is loaded. Its probe function is called with a pointer to the platform device. The driver registers a v4l2_device and creates
I2C subdevices using the data supplied through platform data.

vpfe capture does similar things.. The vpfe capture is already part of the v4l2 sub system. You can find the code under /drivers/media/video/davinci

Quoted 5 lines

3. The omap34xxcam driver calls the ISP core with a pointer to the v4l2_device
structure to register all ISP subdevices. The ISP core maintains pointers to
the ISP subdevices.

I was thinking in similar lines. We could define configuration for the ISP sub devices in the platform code. For example, register IO space for each IP (Preview Engine, Resizer, H3A etc) can be configured at platform level. But how is the sub device operations invoked from the camera driver? Are there wrapper functions in ISP core for each of the operations?

Quoted 6 lines

As ISP submodules (CCDC, previewer, resizer) are not truly independent, we were not planning to split them into separate kernel modules. The ISP core needs to call explicitly into the submodules for instance to dispatch interrupts.

IMO, to re-use, we need to define common operations across all of the
ISP sub modules (ccdc ops, resizer ops, preview-ops etc) since at hight
level hardware provide similar functions . I am assuming ISP core is a glue layer that load up all of the ISP sub devices and provide function calls to set format, get format, etc, right? Currently vpfe capture is interfaced to ccdc modules through an interface (ccdc_hw_if.h). I am planning to
convert it to a sub device. In that case, ccdc along with other sub modules
will be loaded up by a vpfe core similar to ISP core. So by standardizing on
common operations, we could re-use the IP (sub device) across VPFE and OMAP.
I will investigate the OMAP tree that Sakari maintains before starting my
development. Also the Resizer and Preview Engine will have to be used for
memory to memory operation as well. So re-use from that perspective is needed as well. There was an RFC for this from Samsung recently. Did you
have a chance to review this?

Quoted 7 lines

It should be possible to use a single CCDC code base across multiple platforms. The ISP core module would depend on the CCDC module directly. I'm
not sure how the CCDC module should be called though. An omap prefix won't work, as it's used on Davinci as well, and an ISP prefix is too generic. Do you have any internal code name for the ISP device used on Davinci and OMAP platforms ?

The ISP is called VPFE in DaVinci. In our internal release, we call it imp
(short for image pipeline). We use the name dm6446_ccdc and dm355_ccdc for
vpfe capture drivers. Different hardware name for the same IP has put us into this trouble :). omap34xx_dm644x_ccdc.c is a big name to consider. Or simply choose one name omap_34xx.c or dm644x_ccdc.c and document the IP's supported in the KConfig config variable description as well driver file header.

Quoted 7 lines

The code will be pushed to linux-omap when available, but I can't commit to any deadline. If you start working on the Davinci driver before we post anything, could you please post patches and/or RFCs on the linux-media mailing
list (and CC'ing us) ? Confronting our ideas as soon as possible will (hopefully) avoid diverging too much in our implementations.

Yes. will do.

Quoted 5 lines

--
Regards,

Laurent Pinchart

#5 lore raw
In reply to Laurent Pinchart
Quoted 18 lines

-----Original Message-----
From: [email protected] [mailto:linux-media- [email protected]] On Behalf Of Laurent Pinchart
Sent: Monday, November 09, 2009 7:36 PM
To: Karicheri, Muralidharan
Cc: Sakari Ailus; [email protected]; Hans Verkuil; Cohen David Abraham; Koskipää Antti Jussi Petteri; Toivonen Tuukka Olli Artturi; Zutshi Vimarsh (Nokia-D-MSW/Helsinki); [email protected]; Aguirre, Sergio; Ivan Ivanov; Stan Varbanov; Valeri Ivanov; Atanas Filipov
Subject: Re: OMAP 3 ISP and N900 sensor driver update

Hi Murali,

Quoted 2 lines · Muralidharan

Sakari,

Quoted 8 lines
Quoted 3 lines

Thanks for the update...

Who is working on the CCDC driver for OMAP35xx?

Just for the sake of correctness, we're working on an OMAP34xx, not an
OMAP35xx. I don't think that makes a big difference though.

[Hiremath, Vaibhav] OMAP3430 and OMAP3530 are exactly same devices.

Quoted 4 lines

As far as I know nobody on our side is currently working on the CCDC driver.
We're focusing on the previewer and resizer first.

[Hiremath, Vaibhav] I believe I should be able to see the current development activity on sakari's repo, right?

Thanks,
Vaibhav

Quoted 94 lines
Quoted 1 line

After a week or so, I need to start migrating the CCDC driver to

sub device

Quoted 1 line

interface so that application can directly configure the

parameters with out

Quoted 1 line

having to go through video node. Ultimately it is expected that

ccdc will

Quoted 1 line

have a device node that will allow application to open the device

and

Quoted 1 line

configure the parameters (in the context of Media controller). But

to begin

Quoted 1 line

with I intend to port the existing CCDC driver for DM6446 and

DM355 to sub

Quoted 1 line

device interface. Since the VPFE IPs are common across DM6446 &

OMAP 35xx,

Quoted 1 line

we could use a common sub device across both platforms.

Coordinating our efforts on that front would indeed be very nice.

Quoted 1 line

So I this context, could you please update me on the CCDC

development

Quoted 1 line

on OMAP platform that you work?

I haven't checked the Davinci VPFE drivers recently. I suppose they already
use the v4l2_subdev interface for their I2C sensors and tuners. If not, that
would be the first step.

On the OMAP34xx platform, the ISP driver is already somehow separated from the
omap34xxcam driver, although not nicely. In a nutshell, here's the current
plan. Parts 1 and 2 are already implemented and code is available in Sakari's
linux-omap tree.

1. Board code registers an omap34xxcam platform device. The platform data
contains an array of v4l2_subdev_i2c_board_info structures filled with
information about all I2C sub-devices (sensor, flash controller, lens
controller, ...).

2. The omap34xxcam driver is loaded. Its probe function is called with a
pointer to the platform device. The driver registers a v4l2_device and creates
I2C subdevices using the data supplied through platform data.

3. The omap34xxcam driver calls the ISP core with a pointer to the v4l2_device
structure to register all ISP subdevices. The ISP core maintains pointers to
the ISP subdevices.

As ISP submodules (CCDC, previewer, resizer) are not truly independent, we
were not planning to split them into separate kernel modules. The ISP core
needs to call explicitly into the submodules for instance to dispatch
interrupts.

It should be possible to use a single CCDC code base across multiple platforms. The ISP core module would depend on the CCDC module directly. I'm
not sure how the CCDC module should be called though. An omap prefix won't
work, as it's used on Davinci as well, and an ISP prefix is too generic. Do
you have any internal code name for the ISP device used on Davinci and OMAP
platforms ?

The code will be pushed to linux-omap when available, but I can't commit to
any deadline. If you start working on the Davinci driver before we post
anything, could you please post patches and/or RFCs on the linux- media mailing
list (and CC'ing us) ? Confronting our ideas as soon as possible will
(hopefully) avoid diverging too much in our implementations.

--
Regards,

Laurent Pinchart
--
To unsubscribe from this list: send the line "unsubscribe linux- media" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#6 lore raw
Quoted 5 lines · Vaibhav
Quoted 4 lines

As far as I know nobody on our side is currently working on the CCDC driver.
We're focusing on the previewer and resizer first.

[Hiremath, Vaibhav] I believe I should be able to see the current development activity on sakari's repo, right?

That's right. Just use the cont branch if you want to pull. Then the log will be garbage. The alternative is to use the devel branch but then you cannot pull, instead use fetch always and manually reapply local changes. Pick your poison. :-)

Signature · 2 lines
Sakari Ailus
[email protected]
#7 lore raw
In reply to Sakari Ailus

Sakari,

Quoted 19 lines

-----Original Message-----
From: Sakari Ailus [mailto:[email protected]]
Sent: Friday, November 06, 2009 6:52 AM
To: [email protected]
Cc: Laurent Pinchart; Hans Verkuil; Cohen David Abraham; Koskipää Antti Jussi Petteri; Toivonen Tuukka Olli Artturi; Zutshi Vimarsh (Nokia-D-MSW/Helsinki); [email protected]; Aguirre, Sergio; Ivan Ivanov; Stan Varbanov; Valeri Ivanov; Atanas Filipov
Subject: OMAP 3 ISP and N900 sensor driver update

Hi,

I have updated the OMAP 3 ISP driver in Gitorious:

<URL:http://www.gitorious.org/omap3camera>

Major changes since the last update:

Quoted 8 lines

- The Nokia N900 (aka rx-51) sensor drivers are available (will be posted to the list shortly)
- Say goodbye to v4l2-int-device, welcome the v4l2_subdevice interface
(thanks to Laurent Pinchart)
- Miscellaneous stability fixes and cleanups
- H3A rework (by David Cohen)
- Resizer rework (by Antti Koskipää)

Does the v4l2_subdevice conversion had some functional regressions?

Or everything is in place still?

Regards,
Sergio

Quoted 10 lines

The next task is then the moving to Media controller, I guess.

Cheers,

Signature · 2 lines
Sakari Ailus
[email protected]
#8 lore raw
Quoted 1 line · Sergio

Sakari,

Hi, Sergio!

Quoted 11 lines
Quoted 7 lines

- The Nokia N900 (aka rx-51) sensor drivers are available (will be posted to the list shortly)
- Say goodbye to v4l2-int-device, welcome the v4l2_subdevice interface (thanks to Laurent Pinchart)
- Miscellaneous stability fixes and cleanups
- H3A rework (by David Cohen)
- Resizer rework (by Antti Koskipää)

Does the v4l2_subdevice conversion had some functional regressions?

Or everything is in place still?

I'd guess more or less so. I'm not aware of any regressions at least. :)

Signature · 2 lines
Sakari Ailus
[email protected]